def knn_classifier():
# Creating model:
model = KNeighborsClassifier()
# Creating an instance of the SklearnClassification TestHarnessModel subclass
th_model = SklearnClassification(model=model, model_author='Jan',
model_description="K-NN classifier, sklearn default params")
return th_model
def support_vector_classifier():
svc = SVC(probability=True)
# Creating an instance of the SklearnClassification TestHarnessModel subclass
th_model = SklearnClassification(model=svc,
model_author='Hamed',
model_description="SVC")
return th_model
def linear_SVC():
lin_svc = SVC(kernel='linear', probability=True)
# Creating an instance of the SklearnClassification TestHarnessModel subclass
th_model = SklearnClassification(model=lin_svc,
model_author='Hamed',
model_description="Linear SVC")
return th_model
def gmm_classification():
# Creating model:
model = GaussianMixture(n_components=2)
# Creating an instance of the SklearnClassification TestHarnessModel subclass
th_model = SklearnClassification(
model=model,
model_author='Jan',
model_description="Gaussian Mixture Classifier, n_components=2")
return th_model
def rocklins_logistic_classifier():
rocklins_logistic_model = LogisticRegression(penalty='l1',
C=0.1,
n_jobs=-1,
solver='liblinear')
th_model = SklearnClassification(
model=rocklins_logistic_model,
model_author='Hamed',
model_description="Rocklin Logistic: penalty='l1' and C=0.1")
return th_model
def svm_classification():
# Creating model:
model = SVC(probability=True)
# Creating an instance of the SklearnClassification TestHarnessModel subclass
th_model = SklearnClassification(
model=model,
model_author='Jan',
model_description="Support Vector Classifier, sklearn default params")
return th_model
def decision_tree_classification():
# Creating model:
model = DecisionTreeClassifier()
# Creating an instance of the SklearnClassification TestHarnessModel subclass
th_model = SklearnClassification(
model=model,
model_author='Jan',
model_description="Decision Tree Classifier, sklearn default params")
return th_model
def naive_bayes_classification():
# Creating model:
nbc = GaussianNB()
# Creating an instance of the SklearnClassification TestHarnessModel subclass
th_model = SklearnClassification(
model=nbc,
model_author='Jan',
model_description="Naive Bayes Classifier, sklearn default params")
return th_model
def baseline_prior():
strategy = 'prior'
baseline_model = DummyClassifier(strategy=strategy, random_state=5)
th_model = SklearnClassification(
model=baseline_model,
model_author="Hamed",
model_description="Baseline DummyClassifier with strategy='{}'".format(
strategy))
return th_model
def rfc_terra(n_estimators=500, n_jobs=-1):
# Creating an sklearn random forest classification model:
rfc = RandomForestClassifier(n_estimators=n_estimators, n_jobs=n_jobs)
# Creating an instance of the SklearnClassification TestHarnessModel subclass
th_model = SklearnClassification(
model=rfc,
model_author='Hamed',
model_description="Random Forest: n_estimators={0}, n_jobs={1}".format(
n_estimators, n_jobs))
return th_model
def gbc_classification():
# Creating model:
model = GradientBoostingClassifier()
# Creating an instance of the SklearnClassification TestHarnessModel subclass
th_model = SklearnClassification(
model=model,
model_author='Jan',
model_description="Gradient Boosting Classifier, sklearn default params"
)
return th_model
def support_vector_radial_basis_classifier():
# RBF SVM hyperparameters optimized by Zhi Li for perovskite data classification
support_vector_model = SVC(C=100000,
gamma=0.1,
cache_size=5000,
max_iter=-1,
kernel='rbf',
decision_function_shape='ovr',
probability=True)
th_model = SklearnClassification(
model=support_vector_model,
model_author="Zhi",
model_description="svm with radial basis function")
return th_model
def gradient_boosted_tree():
# Gradient Boosted tree with hyperparams configured on stateset 0021
xgb = GradientBoostingClassifier(
learning_rate= 1,
loss='deviance',
max_depth= 10,
max_features='auto',
n_estimators=100,
)
th_model = SklearnClassification(model=xgb,
model_author="Scott Novotney",
model_description="gradient boosted tree with n_trees=10,max_depth=5")
return th_model
def gaussian_naive_bayes_classification():
'''
:param n_estimators:
:param max_features:
:param criterion:
:param min_samples_leaf:
:param n_jobs:
:param class_weight:
:return:
'''
# Creating an sklearn random forest classification model:
gnb = GaussianNB()
# Creating an instance of the SklearnClassification TestHarnessModel subclass
th_model = SklearnClassification(model=gnb,
model_author='Mohammed',
model_description="Gaussan Naive Bayes")
return th_model
def random_forest_classification(n_estimators=361,
max_features='auto',
criterion='entropy',
min_samples_leaf=13,
n_jobs=-1,
class_weight="balanced"):
# Creating an sklearn random forest classification model:
rfc = RandomForestClassifier(n_estimators=n_estimators,
max_features=max_features,
criterion=criterion,
min_samples_leaf=min_samples_leaf,
n_jobs=n_jobs,
class_weight=class_weight)
# Creating an instance of the SklearnClassification TestHarnessModel subclass
th_model = SklearnClassification(
model=rfc,
model_author='Hamed',
model_description=
"Random Forest: n_estimators={0}, max_features={1}, criterion={2}, min_samples_leaf={3}, n_jobs={4}"
.format(n_estimators, max_features, criterion, min_samples_leaf,
n_jobs))
return th_model
def rxn_only_xgb():
model = RxnOnlyXGB()
return SklearnClassification(model=model,
model_author="Scott Novotney",
model_description="RBF xgb using only 3 rxn features",
)
def rxn_only_svm():
model = RxnOnlySVM()
return SklearnClassification(model=model,
model_author="Scott Novotney",
model_description="RBF svm w/only rxn features",
)
def rxn_intuition_svm():
model = RxnRatioOnlySVM()
return SklearnClassification(model=model,
model_author="Scott Novotney, Josh Schrier",
model_description="RBF xgb w/org/inorg ratio, inorganic and acid amounts ",
)
